Increasingly being accepted as a research-assist tool. In extreme cases, even as co-author. Generally can be classified into 4 categories. 

The apprehension on AI-assisted papers is that AIs hallucinate (they make up things, scientists are still trying to fix this). This is because AIs are incentivized to always generate an answer even if it doesn't know. During training AI models are rewarded for guessing more than admitted uncertainty.

Authors are responsible for what they write. Make sure you fact-check everything before publication. Academic misconducts usually result in heavy penalties in academia.

Horror Stories

Guo et al. — AI-Generated Anatomical Figures, Frontiers in Cell and Developmental Biology, 2024

The paper was authored by three scientists in China, edited by a researcher in India, reviewed by two people from the U.S. and India, and published in the open access journal — yet despite undergoing multiple checks, it was published with AI-generated figures that went viral on social media because of their absurdity. VICE The Midjourney-generated images included anatomically impossible rat anatomy with garbled labels such as "testtomcels" and "iollotte sserotgomar." Frontiers retracted it on grounds the article did not meet standards of editorial and scientific rigour. Crucially, Frontiers' policy explicitly states that "the author is responsible for checking the factual accuracy of any content created by the generative AI technology."

Mata v. Avianca, Inc. — U.S. Federal Court, S.D.N.Y. (2023)

The landmark case. Attorney Steven Schwartz used ChatGPT to draft a legal brief and submitted six entirely fabricated case citations — complete with fake judges, fake quotes, and fake internal citations. When confronted, Schwartz asked ChatGPT directly: "is varghese a real case?" — ChatGPT confirmed it was real and that it "can be found on legal research databases such as Westlaw and LexisNexis." He believed it. CNN The cases did not exist. Judge Castel dismissed the personal injury case against Avianca and ordered the plaintiff's attorneys to pay a $5,000 fine, noting numerous inconsistencies and describing one of the legal analyses as "gibberish." Wikipedia The sanctions were not imposed merely for using ChatGPT — they were imposed for continuing to stand behind the fake citations after being put on notice by opposing counsel and the Court.

Loves AI so much that they allow it to be a co-author. Extreme positive case, controversial, unconventional, invites attention and scrutiny. 

Piece by Nature on this topic.

Recommendation: Risky if AI named as co-author. No.

Allows use AI to assist, but needs to declare what AI is used, what is it used for, and where it is used. This is quite common nowadays and I see it becoming mainstream. Big publishers have AI policy now. 

Example: Elsevier | Wiley | Taylor & Francis

Recommendation: AI-Friendly. Yes

Does not yet have a policy on AI. Will probably follow suit with big publishers and allow in the future.

Recommendation: No rules against it. Yes

Extreme negative. Disallows the use of AI altogether. Just pick another journal. Avoid.

Recommendation: No

AI detectors actually also use AI to detect AI. It's learning what constitutes AI writing, and what doesn't. This line is being blurred quickly now as AI tools are getting better and better at producing quality human-like outputs. 

The AI detector can make mistakes. Show in the side is an example of an epic fail. The Declaration of Independence was flagged as AI generated?

Even TurnItIn, one of the leading plagiarism checkers, is careful about the use of its AI detector. It is disabled by default, and if you turn it on, it warns that its results may be inaccurate to avoid legal reprecussions.

Their official sources state: "Our AI writing detection model may not always be accurate (it may misidentify human-written, AI-generated, and AI-generated and AI-paraphrased text), so it should not be used as the sole basis for adverse actions against a student. It takes further scrutiny and human judgment in conjunction with an organization's application of its specific academic policies to determine whether any academic misconduct has occurred. Turnitin does not make a determination of misconduct, rather it provides data for educators to make an informed decision. The percentage on the AI writing indicator should not be used as the sole basis for action or a definitive grading measure."